Hitachi Storage Virtualization

8/7/2019 Hitachi Storage Virtualization

1/74

Broughttoyouby

HitachiDataSystemsEdition Maximize storage

utilization and

simplify management

Storage

Virtualization

Adrian De Luca

Mandar Bhide

A Referencefor the

Rest of Us!FREE eTips at dummies.com


2/74

These materials are the copyright of Wiley Publishing, Inc. and anydissemination, distribution, or unauthorized use is strictly prohibited.


3/74

by Adrian De Luca and Mandar Bhide

StorageVirtualization

FOR

DUMmIES

HITACHI DATA SYSTEMS EDITION



4/74

Storage Virtualization For Dummies, Hitachi Data Systems Edition

Published byWiley Publishing, Inc.111 River StreetHoboken, NJ 07030-5774

Copyright 2010 by Wiley Publishing, Inc., Indianapolis, Indiana

Published by Wiley Publishing, Inc., Indianapolis, Indiana

No part of this publication may be reproduced, stored in a retrieval system or transmitted in anyform or by any means, electronic, mechanical, photocopying, recording, scanning or otherwise,except as permitted under Sections 107 or 108 of the 1976 United States Copyright Act, withoutthe prior written permission of the Publisher. Requests to the Publisher for permission should beaddressed to the Permissions Department, John Wiley & Sons, Inc., 111 River Street, Hoboken, NJ07030, (201) 748-6011, fax (201) 748-6008, or online at http://www.wiley.com/go/permissions.

Trademarks: Wiley, the Wiley Publishing logo, For Dummies, the Dummies Man logo, A Referencefor the Rest of Us!, The Dummies Way, Dummies.com, Making Everything Easier, and relatedtrade dress are trademarks or registered trademarks of John Wiley & Sons, Inc. and/or its affili-

ates in the United States and other countries, and may not be used without written permission.All other trademarks are the property of their respective owners. Wiley Publishing, Inc., is notassociated with any product or vendor mentioned in this book.

LIMIT OF LIABILITY/DISCLAIMER OF WARRANTY: THE PUBLISHER AND THE AUTHORMAKE NO REPRESENTATIONS OR WARRANTIES WITH RESPECT TO THE ACCURACY ORCOMPLETENESS OF THE CONTENTS OF THIS WORK AND SPECIFICALLY DISCLAIM ALLWARRANTIES, INCLUDING WITHOUT LIMITATION WARRANTIES OF FITNESS FOR APARTICULAR PURPOSE. NO WARRANTY MAY BE CREATED OR EXTENDED BY SALES ORPROMOTIONAL MATERIALS. THE ADVICE AND STRATEGIES CONTAINED HEREIN MAYNOT BE SUITABLE FOR EVERY SITUATION. THIS WORK IS SOLD WITH THE UNDER-STANDING THAT THE PUBLISHER IS NOT ENGAGED IN RENDERING LEGAL, ACCOUNT-ING, OR OTHER PROFESSIONAL SERVICES. IF PROFESSIONAL ASSISTANCE IS REQUIRED,

THE SERVICES OF A COMPETENT PROFESSIONAL PERSON SHOULD BE SOUGHT.NEITHER THE PUBLISHER NOR THE AUTHOR SHALL BE LIABLE FOR DAMAGES ARISINGHEREFROM. THE FACT THAT AN ORGANIZATION OR WEBSITE IS REFERRED TO IN THISWORK AS A CITATION AND/OR A POTENTIAL SOURCE OF FURTHER INFORMATIONDOES NOT MEAN THAT THE AUTHOR OR THE PUBLISHER ENDORSES THE INFORMA-TION THE ORGANIZATION OR WEBSITE MAY PROVIDE OR RECOMMENDATIONS IT MAYMAKE. FURTHER, READERS SHOULD BE AWARE THAT INTERNET WEBSITES LISTED INTHIS WORK MAY HAVE CHANGED OR DISAPPEARED BETWEEN WHEN THIS WORK WASWRITTEN AND WHEN IT IS READ.

ISBN: 978-0-470-59770-5

Manufactured in the United States of America

10 9 8 7 6 5 4 3 2 1



5/74

About the AuthorsAdrian De Lucabrings over 15 years of experience ininformation technology to Hitachi Data Systems. In his role asDirector of Storage Management & Data Protection for AsiaPacific, he is responsible for developing and managing thestorage solutions business in the region, with a clear focuson storage management, virtualization, and data protectionsolutions. Adrian has maintained senior positions in systemsengineering and software development. He holds a Bachelor ofApplied Science in Computer Science from the Royal Melbourne

Institute of Technology in Australia.

Mandar Bhide has over 18 years of experience in informationtechnology. In his role with Hitachi Data Systems, as TechnicalDirector of Storage Management & Tiered Storage Solutions forAsia Pacific, he is responsible for providing technical supportfor the storage solutions business and sales development, with aclear focus on storage management and virtualization solutions.Mandar has maintained positions in product management andsoftware quality assurance in his previous roles. He holds aBachelor of Computer Science and Masters in ManagementScience from Pune University in India.

Hubert Yoshida(author of the Foreword) is Vice Presidentand Chief Technology Officer of Hitachi Data Systems. In 2006,he was named CTO of the Year by Jon Toigo and, in October2006, Byte and Switch named him one of Storage NetworkingsHeaviest Hitters. His blog was recently ranked among the topten most influential within the storage industry by Network

World.

Hu is a graduate of the University of California at Berkeley witha degree in Mathematics. He has authored several papers onstorage area networks, fibre channel, multi-protocol SANs, andstorage virtualization technologies. Hu currently sits on theScientific Advisory Board for the Data Storage Institute of theGovernment of Singapore.

Authors AcknowledgmentsAdrian De Luca and Mandar Bhide wish to thank Hitachi DataSystems Hubert Yoshida, David Merrill, and Ojay Bahra for theirassistance with the writing of this book.



6/74

Publishers Acknowledgments

Were proud of this book; please send us your comments through our Dummies

online registration form located at http://dummies.custhelp.com. For othercomments, please contact our Customer Care Department within the U.S. at877-762-2974, outside the U.S. at 317-572-3993, or fax 317-572-4002. For details onhow to create a custom For Dummies book for your business or organization,contact [email protected]. For information about licensing the For Dummiesbrand for products or services, contact BrandedRights&[email protected].

Some of the people who helped bring this book to market include the following:

Acquisitions, Editorial, andMedia Development

Senior Project Editor:Zo Wykes

Australian Edition Editor:Kerry Davies

Editorial Manager: Rev Mengle

Business DevelopmentRepresentative:Karen Hattan

Custom Publishing ProjectSpecialist: Michael Sullivan

Production

Proofreader: Marguerite Thomas

Publishing and Editorial for Technology Dummies

Richard Swadley, Vice President and Executive Group Publisher

Andy Cummings, Vice President and Publisher

Mary Bednarek, Executive Director, Acquisitions

Mary C. Corder, Editorial Director

Publishing and Editorial for Consumer Dummies

Diane Graves Steele, Vice President and Publisher, Consumer Dummies

Composition Services

Debbie Stailey, Director of Composition Services



7/74

Table of ContentsIntroduction .......................................................1

Whats in This Book ...................................................................2

Icons Used in This Book ............................................................3

Chapter 1: Storage Virtualization: Getting Started . . . . . . . . . 5

Defining Virtualization ...............................................................5Looking for What Characterizes an Ideal System ..................7

Storage Virtualization: A Stroll Down

Memory Lane...........................................................................7

Drivers for Storage Virtualization ..........................................10

Out of storage again! ......................................................11

Reducing cost .................................................................12

Managing service levels ................................................12

Adding agility and flexibility .........................................13

Repurposing and adding value toexisting assets .............................................................14

Going green ..................................................................15

Chapter 2: Types of Storage Virtualization . . . . . . . . . . . . . . . 17

Storage Virtualization in Three Layers of

Infrastructure ........................................................................19

In the server ....................................................................19

In the storage network ..................................................22

In the storage controller ...............................................26Choosing the Right Solution ...................................................29

What are your objectives? ............................................29

Where are you today? ....................................................29

Price versus cost ............................................................30

Matching growth to scalability .....................................30

Can you afford an outage? ............................................30

Chapter 3: The Economics of Storage

Virtualization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31How Do You Measure the EconomicBenefits? .................................................................................32

So Where Are the Savings in Storage? ...................................34

Where Do You Start? ................................................................35

Hitachis Storage Economics Model .......................................36



8/74

Storage Virtualization For Dummiesvi

Chapter 4: Making Storage Virtualization Real. . . . . . . . . . . 39

Key Concepts in Storage Virtualization .................................40

Planning for the Virtualized Storage Environment ..............41

Implementing Storage Virtualization .....................................42

Partitioning of storage hardware .................................43

Internal and external storage configurations .............43

Migrating Data to a Virtualized Storage

Environment ..........................................................................45

Classifying the data types .............................................46

Creating storage tiers ....................................................46

Uninterrupted data migration ......................................47

Chapter 5: Getting the Best from StorageVirtualization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49

Optimizing Performance and Improving

Storage Utilization ................................................................49

Understanding thin, or dynamic,

provisioning .................................................................50

Creating a dynamic provisioning pool ........................50

Benefits of dynamic provisioning ................................51

Management Challenges in the Virtualized

Storage Environment ............................................................53

Dealing with the challenges ..........................................55

Managing performance and capacity ..........................56

Chapter 6: Ten Best Practices for Deploying StorageVirtualization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57

#1: Choose the Right Storage for

the Hosts Mapping ...............................................................57

#2: Assign External Storage Configurations ..........................58

#3: Format the Virtual Disk .....................................................58

#4: Choose the Right Disk Attributes for

Data Migration .......................................................................59

#5: Migrate the Data .................................................................59

#6: Classify the Data .................................................................60

#7: Understand the Workload or I/O Profile .........................61

#8: Design the Dynamic Provisioning

Storage Pool ...........................................................................62

#9: Create a Dynamic Provisioning Pool ...............................63

#10: Plan and Design a Virtualization Project ......................63



9/74

Foreword

When I was a young man, fresh out of college, I couldthrow all my worldly possessions into the back ofmy Chevrolet and go wherever I needed to be. Since then, Igot married, had a family, bought a house, bought furniture,

appliances, and tons of other stuff to support my familyslifestyle, and now it would take several truckloads to move mewith all my stuff. I have lived in the same community for the past20 years and have built a life here. It would be difficult for meto move if my work required it. However, with the availability ofthe Internet and broadband, my workplace has been virtualizedand I can do my work without disrupting my life and moving allmy stuff.

Just like my life, businesses are burdened with tons of stuff,

mainly data, that they must store, preserve, protect, and access.That burden is nearly doubling every year and is beginning tostrangle businesses, limiting their ability to do their work.

Data center managers are handling aging applications that arerunning on aging infrastructures and facilities that were builtwhen electricity was readily available and carbon footprints werenot a concern. They spend more and more of their data centerbudget on maintaining old systems instead of investing in new

systems that are more efficient and environmentally friendly. Thereason it is so difficult to modernize IT is the burden of data thatmust be carried forward with the modernization of applicationsand infrastructures. Ten years ago, before the turn of themillennium, changes could be made by suspending operationsduring off hours while data was migrated or transitioned to anew storage platform. Storage frames were directly attached toapplication servers and capacities were small enough that theycould be moved within a day.

Since then, the explosion of data, globalization, and onlineaccess requires applications to be available around the clock,and suspension of operations must be kept to an absoluteminimum. The capacity of storage frames has increaseda hundredfold and many more applications now shareaccess to these large storage frames, making it difficult tocoordinate downtime for shared storage frames. This makes



10/74

Storage Virtualization For Dummiesviii

it nearly impossible to migrate applications or modernize theinfrastructure without a major outage.

In my case, my ability to work was greatly simplified throughvirtualization of the workplace. I can do my work anywhere thereis broadband access.

In the business world, there is a similar need to virtualizestorage in order to simplify operations and make it easy tomodernize IT. Only through virtualization of storage can datacenters hope to have the agility to modernize IT without majordisruptions to their business.

To many people, the subject of storage virtualization is verycomplex. The purpose of this book is to simplify this complexity,remove the misconceptions and hype around storagevirtualization, and leave you with a basic understanding of whatstorage virtualization is and what it can do for you.

Hubert YoshidaVice President and Chief Technology Officer

Hitachi Data Systems



11/74

Introduction

Alot has been written, taught, and even debated aboutvirtualization in the information technology industry overthe past few years, especially in the server world where it is bestknown. Vendors like VMware and Microsoft actively promote

how their virtualization software provides efficient resourceutilization, dynamic capabilities and ease of management, whilethe hardware vendors fiercely contend to be the company thathas the lowest power consumption, better interoperability, andease of deployment and support.

Vendor marchitecture aside, all organizations are looking toimprove efficiency, drive up productivity, and maximize everydollar of their IT investments. Virtualization has become aneffective way to do more with less, especially in these tough

economic times.

However, for all the hype generated over virtualization, thereremains a considerable number of misconceptions and generalconfusion. Many people think virtualization is some new whiz-bang technology when, in fact, the concept has been around along time, with its origins dating back to the mighty mainframeera. The philosophy of aggregating physical components in acoordinated way to increase systems utilization and efficiency is

very mature and has been successfully applied in many areas ofIT beyond servers, such as networking and applications.

So what about storage virtualization? Can the same benefitsbe realized when virtualization is applied to the storage areanetwork (SAN)? The fact is that virtualization in the storageenvironment shares many of the same benefits of servervirtualization and, most importantly, is highly complementary,extending agility and cost savings to the data assets.

As a reader of this book, you are most likely looking tounderstand what virtualization is in the storage world, whatbenefits you are likely to realize, and how best to deployvirtualization in your environment to get the most out of it.

Storage Virtualization For Dummies, Hitachi Data Systems Edition,explains the differences in implementation among variousvendors and, more specifically, how it is applied by Hitachi.



12/74

Storage Virtualization For Dummies2

It also tries to dispel the myths to bring an objective look at howthis technology is best applied.

Much like other forms of virtualization, the path is not alwayssimple. Objectives need to be clearly understood and agreedat the outset, and a sobering understanding of where you aretoday needs to be discovered. In addition, as with any IT project,preparation and planning needs to be invested in the design, andthe solution correctly implemented, if you are to realize the truebenefits of virtualization.

The good news is that this journey does not need to be walkedalone. A lot of experience has been developed over the five or soyears during which this technology has made its breakthroughinto storage. Many organizations have transitioned theirdysfunctional legacy storage environments into highly flexibleand agile infrastructures that more closely align and adaptto their business needs. Vendors have the technology, bestpractices, and skills to assist you, and resources such as thisbook help put you on the right track to succeed.

Whats in This BookThis book comprises six chapters.

Chapter 1 sets the framework of storage virtualization, givingyou a summary of what it is and how you apply it, as well asan outline of how it came about. This chapter also analyzesthe drivers of storage virtualization (why youll even think of

applying it).

In Chapter 2 we talk about the types of storage virtualizationyou can choose from and how to select the right type for yoursituation. Chapter 3 looks at the economics of storage, includingan outline of Hitachis Storage Economics model and a casestudy.

Chapter 4 delves into more detail on planning and implementing

storage virtualization according to your own objectives.This chapter also examines how to migrate your data to thevirtualized storage environment. And in Chapter 5 we explainhow to optimize and manage the systems performance.



13/74

Introduction 3

Last but not least, noFor Dummies book would be completewithout a chapter of tens. So Chapter 6 quite succinctly

explains the ten best practices for physical configurations, datamigrations, and thin, or dynamic, provisioning, and tips fordesigning your first storage virtualization project.

Icons Used in This BookThroughout this book we use a series of icons in the marginsthat help flag special information. Heres what to look for.

When we tell you something about storage virtualization thatbears remembering, we mark it with the Remember icon.Descriptions that appear beside this icon are worth storing awaybecause they help build your understanding.

This flags a shortcut or easy way to do something.

This icon indicates the information is really for the propellerspinners; you dont need to know all about how this works butyou may find it interesting.

This icon indicates a resource on the Internet you can look at toget further information.



14/74




15/74


16/74


Abstracting the physical components of computing resourcesoptimizes the way in which other systems, applications, or end

users interact with them. This definition summarizes two veryimportant characteristics of virtualization: It should reduce thecomplexity when compared with managing devices discretely, aswell as add greater capability to improve services a bit like aone-plus-one-equals-three equation.

When described in terms of storage, the Storage NetworkingIndustry Association (SNIA), the independent body of thestorage industry, has a more specific definition (two of them,in fact) for storage virtualization, from its technical tutorial onvirtualization:

1. The act of abstracting, hiding, or isolating the internalfunctions of a storage (sub)system or service fromapplications, computer servers, or general networkresources for the purposes of enabling application andnetwork independent management of storage or data.

2. The application of virtualization to storage services or

devices for the purpose of aggregating, hiding complexity,or adding new capabilities to lower-level storage resources.

Put simply, storage virtualization aggregates storagecomponents, such as disks, controllers, and storage networks,in a coordinated way to share them more efficiently amongthe applications it serves. Pooling these resources in a logicalway helps remove the physical barriers that would otherwiseexist and maximize the full potential of these resources. Hidingthe complexity associated with all the components required to

deliver storage helps simplify the operations of managing theenvironment and, more importantly, provides greater flexibilityto meet the needs of applications.

Think of virtualization in the context of using a telephone.When you pick up the handset or initiate a call on your mobilephone, as the caller you are completely unaware of how you areconnected to the other party. The complexity of establishingcommunication is completely invisible; you are oblivious as to

the number of hops across the telephone network or the typeof communication involved (such as satellite or undergroundcable), or even if there is a failure in the network. This isvirtualization technology at work, using available resources asefficiently and intelligently as possible to deliver a service.



17/74

Chapter 1: Storage Virtualization: Getting Started 7

Looking for What Characterizesan Ideal System

Many vendors offer storage virtualization solutions, eachimplemented in a different way or offered in different packaging.Therefore, it is important to understand the characteristics of anideal storage virtualization solution.

A good storage virtualization solution should:

Enhance the storage resources it is virtualizing through theaggregation of services to increase the return of existingassets.

Not add another level of complexity in configuration andmanagement.

Improve performance rather than act as a bottleneck inorder for it to be scalable.Scalabilityis the capability of asystem to maintain performance linearly as new resources

(typically hardware) are added.

Provide secure multi-tenancy so that users and data canshare virtual resources without exposure to other usersbad behavior or mistakes.

Not be proprietary, but virtualize other vendor storage inthe same way as its own storage to make the managementseamless.

Storage Virtualization: A StrollDown Memory Lane

With the introduction of commercial computers back in the1950s, like IBMs mainframes, the magnetic disk storage usedto maintain information was directly attached via a cable, or abus, to the computers central processing unit (CPU) and volatile

random access memory (RAM). This method of attachment,commonly referred to as direct attached storage (DAS), provideda modest amount of non-volatile storage to a single computingsystem.

Over the course of the next few decades, standardprotocols (thelanguage used to communicate between devices) emerged, suchas the small computer system interface (SCSI), making it easy to



18/74


connect different vendor storage to the systems. This methodof connectivity extended beyond traditional magnetic storage

to devices such as CD-ROMs, tape drives and autoloaders, andJBOD (just a bunch of disks). Although the different types ofstorage flourished and fault-tolerant designs provided greaterreliability, their connectivity was still confined to single serversor workstations, limiting the utilization of the media.

The first wave of storage virtualization appeared in the cachecontroller arrays. Large numbers of magnetic disks were pooledtogether and arranged in a way to provide fault tolerance toprotect against individual disk failures. This system becameknown asRAID (redundant array of independent disks). Utilizinga common pool of cache memory, applications could workwith a logical image of a data block rather than working withthe actual data block as it spun around on a disk platter. Thisimproved performance by masking the seek and rotation delaysof a mechanical disk. It also enabled mainframes to use lowercost disks.

In the early 1980s, in the midst of the computer network

revolution that allowed servers to talk to one another andexchange information over standard mediums like ethernet,vendors such as Novell and Sun Microsystems figured that, ifservers can share data, then why not share their storage aswell? By harnessing the power of the network, network attachedstorage (or NASas it is commonly known today), allowedservers to treat storage in another room, or potentially anothercity, as though it was right there, physically attached to theserver. This offered unprecedented levels of flexibility to store

data centrally but make it available widely. It also allowedmultiple users to write and read to the storage at the same time,providing universal access to the same data sets and increasingcollaboration between users. As this common way to accessdata became popular, other operating system vendors startedoffering this capability in their systems, and companies likeNetwork Appliance, EMC, and Hitachi Data Systems packaged thefunctionality into appliances for easy deployment and increasedscalability.

The next advancement in storage virtualization came with theintroduction of storage area networks (SANs) in the mid to late1990s, when the concept of storage resource aggregation made itout of the box.



19/74


What is a storage area network?A storage area network (SAN) is anarchitecture to connect detachedcomputer storage devices, such asdisk arrays, tape libraries, and opticaljukeboxes, to servers in a way that thedevices appear as local resources.

SANs deliver storage to servers at ablock level, and feature mapping andsecurity capabilities to ensure onlyone server can access the allocatedstorage at any particular time.

The protocol, or language, used to communicate between storagedevices and servers is SCSI (smallcomputer system interface).

Traditionally, SANs have used opticalconnectivity called fibre channel (FC)due to its gigabit transfer speeds andability to work over long distances.However, more recently ethernet-based storage networks have become

popular due to the ubiquity (and there-fore lower cost) of the technology inthe TCP/IP world (transmission controlprotocol/Internet protocol) and theadoption of the SCSI protocol, callediSCSI (Internet small computer systeminterface) to the medium.

SANs helped organizations consolidate their storage assetsto improve capacity utilization by sharing their storageresource effectively. This simplified management by usingcommon software tools, and enabled replication of criticalinformation over long distances to provide greater levelsof protection against data corruption and disaster events.

Many larger organizations, typically financial institutionsand telecommunication providers, were among the first toimplement SANs.

Even though these improvements helped organizations extractbetter value from their storage investments, a number ofproblems still remained. Islands of SANs were created due topoor interoperability between different vendors and devices,utilization levels were still relatively low due to traditionalstorage allocation practices being used, and the lack of mobilityor replication between different storage vendors meantorganizations were still hamstrung.



20/74


What is network attached storage?Network attached storage (NAS) has

traditionally been a single device thatcontains disk storage and computercomponents (CPU, memory, networkports) and whose sole purpose is todeliver storage to servers. NAS is

sometimes referred to as a file serverdue to its ability to share a commonpool of files with multiple servers.By managing the disks, file systems,and volumes, it optimizes and simpli-fies the delivery of storage to serversin a network environment, and canprovide services such as redundancyand replication, relieving servers of

these tasks.

NAS storage typically connects toan organizations TCP/IP network viaethernet (the same used by servers)and uses specific languages, orprotocols, like NFS (network filesystem) or CIFS (common Internet file

system), based on the server messageblock protocol, to exchange infor-mation between servers.

More recent implementations ofNAS appliances use SANs to accessattached block-based storage.

In response to these challenges, in the early years of the 2000s,storage vendors began to introduce advanced virtualizationfeatures into their products. These features went beyondutilization improvements, also providing external connectivityof heterogeneous storage, non-disruptive data migrationand mobility features, enhanced business continuity, logicalpartitioning, tiered storage, and thin provisioning. With thin

provisioning(explained further in Chapter 5), the physicalcapacity of a volume or file system is allocated whenapplications write data, rather than being pre-allocated at thetime of provisioning. Although the need for these capabilitieswas well understood by vendors, each vendor had a differentphilosophy on where the virtualization should reside.

Drivers for Storage VirtualizationThere is no denying that advancements in computer technologyhave redefined methods of communication, creativity, andbusiness management. This revolution has created an explosionof information generated for both businesses and home users.In the business world, applications that manage emails and Web



21/74


sites, as well as customer relationship management (CRM) andsales systems, are all critical to keeping the doors open. If these

applications are unavailable for whatever reason, businessescant service their customers and theyll go elsewhere. TheWeb has emerged as a source of rich media (images, video,and audio) and continues to grow unabated, with serviceslike Facebook, Flickr, and YouTube creating unprecedentedcommunities and collaboration between people all over theworld. This exponential growth, together with the need to accessinformation, has placed huge demands on the IT staff lookingafter them, especially on their storage infrastructures.

With this explosion of new applications and services, the valueand importance of the information generated has become morevital. Business intelligence and decision support systems mineinformation in databases to create valuable knowledge that helpsorganizations make prudent decisions and stay competitive.Also, with regulatory compliance coming into play, the need toretain and protect corporate data for predetermined periodsof time is now inscribed in law. Similarly, the accessibility ofinformation and media from the Internet and Web sites, in

particular its availability, determines how successful a businessis. If users cant get what they want, when they want it, fromtheir browser in a couple of seconds, then they simply goelsewhere.

All this creates several challenges in the storage environment.

Out of storage again!Perhaps the most nerve-wracking part of a storageadministrators job is when he has to go cap in hand to theIT manager or CIO and ask for money to buy more storage. Likeneeding to ask your parents for more pocket money, and havingto first explain where your weekly allowance went, systemadministrators face a similar judgment. Unfortunately, explainingwho took all the storage is not necessarily simple, it just seemsto vanish!

The main reason this happens is applications are typicallyallocated more storage than they need. Application and systemadministrators often request as much storage as they can get,rather than what theyre likely to need, since it is a scarceresource. Twelve months down the track they find they haveused only a small portion of their allocation, say 30 percent.Due to traditional storage provisioning methods, the unused



22/74


70 percent remains stranded. Although it hasnt been usedby the application and is free, it cant be allocated to other

applications as it is dedicated to the original server. Multiply thissituation out to tens or hundreds of servers and applicationsand its not hard to see how this inefficiency quickly adds up.

Compounding the problem is the common practice of creatingmultiple copies of data. For every byte of production data,organizations create any number of replicas for backup andrecovery, disaster recovery, data mining, development andtesting. Many organizations exposed to industry or governmentregulations retain several more copies for compliance or otherrecords management requirements long after the data hasbecome stale.

Countless surveys and assessments of storage capacityutilization have been conducted over the years, most revealingthat organizations suffer from very poor actual utilization, as lowas 30 to 40 percent.

Reducing costAs organizations deploy more and more applications intotheir environment, with consequent growth of data (oftenexponentially), many find they need to buy more storage morefrequently, putting a real strain on budgets. As more storageis added, without effective storage management tools andprocesses, more IT people are needed to keep it going, andan unending cycle perpetuates. As more allocation requestsare met, more people are needed to monitor the health of theinfrastructure and protect the critical information by ensuringit is replicated and can be recovered in the event of a failure,requiring more storage.

With IT budgets and headcount staying flat or in many situationsactually decreasing, organizations need to find more efficient andautomated ways of managing their valuable information assets inthe face of static or shrinking resources.

Managing service levelsIT systems have become the lifeblood of most organizationsin this age. With the dependency on these applications toservice day-to-day business activities, making sure they deliverconsistent and reliable service is critical.



23/74


Take email, for example. Most employees take this servicefor granted as a fast way to communicate between their

departments and with customers, send out information orquotations, and approve business decisions. If the email serverwere to slow down or be unavailable for a couple of days,not only would employees be unhappy, but the organizationsproductivity would plunge to a standstill. In this highlyconnected, real-time world, having to resort to traditional formsof communication and paper-based processes would significantlyreduce the number of transactions the organization could doand, if the unavailability persisted, could place the business inreal jeopardy. Therefore it is imperative that IT staff monitorthe health of their infrastructure end to end, including storage,as the key piece to delivering dependable service levels to thebusiness.

Adding agility and flexibilityWith such demanding service levels that require applicationsto be available 99.99 percent of the time (or more!), bringing an

application down for servicing and maintenance, or migratingto new server or storage platforms, would be unthinkable. Trytelling the salespeople in your office that you have to shut downtheir quoting and order-entry system for a day and theyre likelyto go after you with pitchforks!

Therefore any such maintenance activities must be donewith the least or, in many cases, no disruption. This poses areal challenge for IT staff as it typically involves coordinatingmultiple people to orchestrate the changes, having soundprocesses to execute the tasks with no data loss, and having fall-back measures in case something goes horribly wrong. All thissounds like a death-defying high-wire circus act reserved only forthe brave. Many organizations try to avoid such activities as itplaces them in a risky situation, but migrations, for example, aresimply unavoidable. Organizations need to look for infrastructurethat is more flexible and agile to cope with these requirements.

In the server world, running out of CPU or memory resources

to meet a given workload is common. With server virtualizationsoftware like VMware, applications and their operating systemscan be seamlessly moved to other physical servers without theapplication skipping a beat.



24/74


For the storage infrastructure, if the storage serving anapplication cant provide adequate performance or availability,

traditionally system and storage administrators would stop theapplication, copy all the data to another location, verify it andhope that the application comes back in good shape. Storagevirtualization technology provides the ability to move data fromone pool of disks to another without any need to do a manualmigration.

Repurposing and adding value

to existing assetsIts funny how things just dont seem to last as long as they usedto. Not so much because of wear and tear but because theybecome superseded by new models with better performance orfeatures.

Most storage subsystems and associated networks costthousands or even millions of dollars. So, with such potentially

short life cycles, its not surprising that organizations are lookingto squeeze every last drop out of their investments.

For many organizations, its not a matter of technology beingsuperseded, but no longer meeting their changing requirements.Most IT departments assess their storage requirementsaccording to the most appropriate features, technology,vendor, and budget to serve their purposes best today.However, business landscapes are constantly transforming, andunforeseeable events can change things in a heartbeat. Certain

applications can become more important, requiring greaterprotection or performance. Your organization could acquireanother company, and you inherit another IT infrastructure tomanage and merge with your own.

The problem is that business and therefore technologyrequirements change rapidly, often within a year, while hardwaremay be capitalized in three to five years. If youve just boughta new storage unit and are only one year into a five-year

capitalization cycle or five-year lease, youre locked out of usingnew storage-based technologies for the rest of the capitalizationor lease period.

When organizations compare the cost of maintenance for theiraging hardware with that of buying new hardware, they areoften motivated to buy new. However, with storage virtualizationsolutions you can breathe new life into existing assets, rather



25/74


than replacing them, and repurpose them for a job with fewerrequirements, or extend functionality, such as thin provisioning

or remote replication (these concepts are explained inChapter 5), to increase their value to the business.

Going greenMore than 20 years of research into carbon pollution nowstrongly suggests that human activity is having a profoundimpact on the Earths climate. The almost indisputable evidencehas now forced many governments to action by implementing

tangible strategies and policies that reduce carbon emissions.As one of the many measures to tackle the problem, it calls forcountries to implement carbon emissions trading schemes tosteadily reduce pollution, but at a cost to businesses.

So what has all this got to do with storage? The net effectof these new trading systems, coupled with the fact thatthe worlds reserves of non-renewable resources are rapidlydepleting, means the cost of power is rising. In the past, the cost

of powering and cooling data centers was negligible and wasoften not even factored into the IT budget. However, accordingto an IDC article in 2008, recent surveys suggest that withcontinued deployment of applications and data doubling every18 months, power and cooling costs will grow eightfold by 2010!In another survey, by Gartner Consulting, findings showed thatthe power consumption required to run and cool data centersaccounts for almost a quarter of global carbon dioxide emissionsfrom the information and communication technology sector.And the StorageIO Group reports that storage itself accounts forbetween 37 and 40 percent of total energy usage from hardware.

With business accounting for a substantial share of greenhousegas pollution, many organizations have developed socialresponsibility policies to help protect the environment.Demonstrating to customers and their respective industries thatthey are not only conscious of the issue, but care about beingpart of the solution, is important for organizations. A numberof large multinational companies, such as Citigroup, Vodafone,

HSBC, and British Telecom, have already committed to carbonreductions within their companies, some aiming to be carbonneutral altogether.



26/74


All these factors are prompting businesses to look at innovativeways to become more efficient users of power in order to

comply with government policies, control their budgets, andprotect their reputations, along with the environment.

Many IT vendors are already actively delivering productsthat have less environmental impact. Through streamliningproduction processes, many vendors have eliminated waste,requiring less carbon-dependent power. Smarter designs ofcomponents such as processors and power supplies, and theability to spin down inactive disks, reduce the running costs, anduse of non-hazardous materials in the product itself protects theenvironment when it is disposed of.

However, these initiatives only go part of the way.

With data storage growing at such exponential rates, togetherwith the inefficiency of traditional storage networking,provisioning and application consumption methods, storagevirtualization can go the next step by eliminating unnecessarywaste.

Consolidating storage though intelligent pooling of storageresources and use of thin provisioning can delay storagepurchases, resulting in far fewer unused disks spinning andconsuming power. By tiering storage resources and aligning theright storage characteristics to the value of the information,organizations can use platforms that utilize disk spin-downtechnologies for archive data, for example, that is infrequentlyaccessed.



27/74

Chapter 2

Types of StorageVirtualization

In This Chapter Examining storage virtualization throughout the infrastructure

Knowing how to select appropriate storage solutions

In todays modern computer architectures, virtualization

exists in almost every layer, from the application to theoperating system, server, networks, and storage devices. Forexample, application clustering technologies such as MicrosoftClusters and Oracles Real Application Clusters (RAC) managethe process of selecting a server to deliver an applicationwithout the user knowing which server its coming from. Servervirtualization allows you to run multiple operating systems onthe same physical hardware platform to improve utilization ofthe central processing unit (CPU) and memory. Most of these

forms of virtualization work together to optimize efficiencythroughout the layers of technology.

This chapter examines the different forms of storagevirtualization and helps you to choose the appropriate type ofstorage virtualization for your environment.

Virtualization can also be implemented in various layers of thestorage infrastructure, starting at the operating systems devicefile through to the host bus adapter, storage network, andstorage array, as shown in Figure 2-1.

Virtualization concepts can also be implemented within thedevices themselves. For example, storage array vendors employvirtualization aspects within devices such as RAID (redundantarray of independent disks). RAID groups together physical



28/74


Figure 2-1: Storage architecture follows a logical system.

hard disks, spreading the data across all the spindles andslicing them up in chunks called LUNs (logical unit numbers)to deliver to servers. This method provides greater protectionagainst a physical drive failure, so if one were to stop workingfor whatever reason, the system would not lose any valuabledata and continue to work. Similarly, storage network vendorsincorporate partitioning capabilities such as zones and virtualstorage area networks (VSANs) to separate workloads formanagement and security purposes.



29/74

Chapter 2: Types of Storage Virtualization 19

Storage Virtualization in ThreeLayers of Infrastructure

Modern storage virtualization technologies pool heterogeneousstorage vendor products together in a specific way to provideadvanced features such as non-disruptive migration of dataand thin provisioning (explained in Chapter 5). This levelof abstraction can be implemented in three layers of theinfrastructure, in the server, in the storage network, and in thestorage controller.

In the serverSome of the earliest forms of storage virtualization came notfrom the storage infrastructure, but from within the server, or,more specifically, the servers operating systems.

With traditional storage hardware devices that connected

directly to servers, the actual magnetic disk was presented toservers and their operating systems as LUNs, where the diskwas arranged into sectors comprised of a number of fixed-size blocks. To allow applications to not only store, but findinformation easily, the operating system arranged these blocksinto a file system. Much like a paper-based filing system, a filesystem is simply a logical way of referencing these blocks intoa series of unique files, each with a meaningful name and typeso they can be easily accessed. For example, take the file nameMy_Summer_Break.doc.My_Summer_Break describes what is

contained within the file and the extension .doc identifies thefile as a document. This naming is a lot more meaningful thanjust a numeric block number.

Although file systems helped to reference information easily, asmore and more of them were created, exhausting the storagespace of the physical LUN, another LUN would be created andgiven to the operating system to continue storing files. To knowwhich data was stored on what LUN, the operating system

would assign each one a volume number, name, or identifier.In Microsoft Windows you are most likely familiar with a lettergiven to each volume such as C:\ or D:\, whereas in UNIXthese look like /dev/hd0 or /dev/hd1. As applications andusers created more files, more volumes were needed to keep up,pretty soon making it very difficult to manage.



30/74


Then operating system vendors came up with the concept ofa logical volume manager (LVM). Much like how file systems

grouped blocks together to present files, LVMs grouped volumesor LUNs together to present larger, more flexible storage poolsto applications, as shown in Figure 2-2. When an LVM started torun out of space, you could concatenate, or add another volume,to make it larger without having to reconfigure the applicationor shut it down. Conversely, if you had a large volume thatyou wanted to slice into smaller chunks, LVMs would let youpartition it in order to separate different information, say oneslice for the operating system itself and the other for user data.

Figure 2-2: Logical volume managers (LVMs) store information in file systems.

Also, since LVMs could hide or abstract where the storagewas actually coming from, you could present multiple physicaldisks and perform a striping operation. The technique of datastripingtakes multiple physical disks, ideally with their owndiscrete path to the server, breaks up the files into even pieces,and spreads the file across multiple storage devices. Writing andreading data in this way reduces the response time and thereby



31/74


increases the performance. All hard disks have a seek time,which is the time it takes for the information being requested on

the spinning disk to arrive under the reading head. By spreadingdata across multiple disks, the process of finding the informationcan be done in parallel, thereby reducing the seek time.

Although striping greatly helped to retrieve data faster, it didpose a significant risk that, if just one of the disks failed, theneffectively the full data could not be retrieved, resulting incorruption. In the 1980s, a new striping method was introduced,RAID, which increased data reliability if disks failed. RAID-1duplicates all the data across every disk, providing a backupcopy for everything. Although this provides high levels ofresilience, it is extremely expensive because you need doublethe amount of storage. RAID-5, shown in Figure 2-3, combinesthe performance of striping plus parity, which provides recoveryin the event of a disk failure. By using a mathematical formulato calculate the parity, if a disk were to fail in the RAID set,then the data can be regenerated and retrieved with no impact.More recently, RAID-6 was introduced to cope with the longregeneration times of large disk drives, writing parity information

to two disks instead of one.

Figure 2-3: How RAID increases resilience to minimize data loss.

Server-based forms of storage virtualization were originallyincorporated into operating systems as software and still remainvery popular today. Here are the key benefits of this approach:

Server-based storage virtualization is highly configurableand flexible since its implemented in the system software.

Because most operating systems incorporate thisfunctionality into their system software, it is very cheap.

It does not require additional hardware in the storageinfrastructure, and works with any devices that can be seenby the operating system.



32/74


A number of downsides to server-based virtualization also exist:

Although it helps maximize the efficiency and resilience ofstorage resources, its optimized on a per-server basis only.

The task of mirroring, striping, and calculating parityrequires additional processing, taking valuable CPU andmemory resources away from the application.

Since every operating system implements file systems andvolume management in different ways, organizations withmultiple IT vendors need to maintain different skill sets andprocesses, with higher costs.

When it comes to the migration or replication of data(either locally or remotely) it becomes difficult to keeptrack of data protection across the entire environment.

Most operating system vendors like Microsoft, IBM, HewlettPackard, and RedHat (Linux) provide at least some capability tovirtualize storage resources. Vendors like Symantec offer moreadvanced forms of server-based storage virtualization.

In the storage networkWith the introduction of network attached storage (NAS) andstorage area networks (SANs) in the late 1990s, it becamepossible to separate disks (and their controllers) from serversand share the storage resources more effectively among allapplications in the IT environment. The storage network becamethe traffic cop for all information being exchanged between

servers and storage devices, and some storage vendors thoughtthis would be the perfect place to manage virtualization.

Network-based storage virtualization embeds the intelligenceof managing the storage resources in the network layer,abstracting the view of real storage resources between theserver and the storage array, either in-band or out-of-band.

The in-bandapproach, sometimes referred to as symmetric,

embeds the virtualization functionality in the I/O (input/output)path between the server and storage array, shown in Figure 2-4,and can be implemented in the SAN switches themselves or inspecialized appliances. All I/O requests, along with the data,pass through the device, with the server interacting with the



33/74


virtualization device, never directly with the storage device. Thevirtualization device analyzes the request, consults its mapping

tables, and, in turn, performs I/O to the storage device. Thesedevices not only translate storage requests but are also able tocache data with their on-board memory, provide metrics on datausage, manage replication services, orchestrate data migration,and implement thin provisioning.

Figure 2-4: In-band network storage virtualization is embedded in the I/O path.

The out-of-bandapproach, sometimes referred to as asymmetric,does not strictly reside in the I/O path like the in-band approach;rather, it works hand in hand with specific virtualization-enabled SAN switches to perform specific look-ups, shown inFigure 2-5. The servers maintain direct interaction with the

storage array through the intelligent switch. The out-of-bandappliance maintains a map (often referred to as meta-data) of allthe storage resources connected in the SAN and instructs theserver where to find it. In this two-step process, the server uses



34/74


special software or an agent, as instructions need to be sentthrough the SAN to make it work. As data never passes through

the virtualization device, performance is only slightly impacted;however, functions such as caching of data are not possible.

Both in-band and out-of-band approaches provide storagevirtualization with the ability to:

Pool heterogeneous vendor storage products in a seamlessaccessible pool.

Perform replication between non-like devices.

Provide a single management interface.

However, only the in-band approach can cache data forincreased performance.

Figure 2-5: Out-of-band network storage virtualization uses intelligent switches

to maintain direct interaction between the servers and the storage arrays.



35/74


Both approaches also suffer from a number of drawbacks:

Implementation can be very complex because the poolingof storage requires the storage extents to be remappedinto virtual extents. This requires a mapping table, whichbecomes a single point of failure, and a vendor lock-in.Migration to a different virtualization solution is extremelydifficult or impossible as all the data must be moved backthrough the mapping table.

The virtualization devices are typically servers runningsystem software and requiring as much maintenance as a

regular server. Clustering is needed to protect the mappingtables and maintain cache consistency between the nodes,which can be risky. Servers are also limited in the numberof storage ports they can support and RAM (random accessmemory) capacity (as storage is limited in cache). As aresult, servers lack scalability.

The I/O can suffer from latency, impacting performance andscalability due to the multiple steps required to completethe request, and limited to the amount of memory and CPU

available in the appliance nodes.

Decoupling the virtualization from the storage once it hasbeen implemented is impossible because all the meta-dataresides in the appliance, thereby making it proprietary.

Solutions on the market only exist for fibre channel (FC)based SANs. These devices are not suitable for Internetprotocol (IP) based SANs, which utilize iSCSI (Internet smallcomputer system interface), NAS, or mainframe servers.

Since both approaches are dependent on the SAN, theyrequire additional switch ports, which involves additionalzoning complexity.

When migrating data between storage systems, thevirtualization appliance must read and write the datathrough the SAN, check status coming back, and maintaina log for any changes during the move that impactperformance.

All of these add complexity and cost to the SAN, whichmakes it very difficult to manage the network withregistered state change notifications (RSCNs, which notifyspecified nodes of any major fabric changes), inter-switchchatter, zoning changes, and buffer credit management.SANs are the third leading cause of application failure afterhuman and software errors.



36/74


The in-band approach has a couple of additional disadvantages:

Specialized software needs to be installed on all servers,making it difficult to maintain.

It also requires specialized SAN switches that support thepropriety virtualization protocol, forcing organizations toreplace existing SAN fabric, which can be expensive.

For all these reasons and more, storage network virtualizationsolutions have had limited traction in the marketplace. Vendorssuch as IBM, EMC, and FalconStor provide network-based

storage virtualization solutions.

In the storage controllerEnterprise-class storage arrays, which have features andcapability suitable for large organizations, have always featuredvirtualization capabilities (some more than others) to enhancethe physical storage resource. One example of this is RAID,for providing data protection from disk failures. Host storagedomains are another example, for virtualizing front-endconnection ports to allow multiple operating systems to use thesame physical port. Many enterprise-class devices incorporatesophisticated switching architectures with multiple physicalconnections to disk drives to provide balanced performanceand resilience, as well as the ability to logically segment internalglobal cache for service quality and security.

With various forms of virtualization already incorporated into

storage controllers for the past 15 years, some storage arrayvendors thought it was logical to extend these capabilities tostorage resources outside the box. Controller-based storagevirtualization was born (see Figure 2-6).

This breed of storage array allows other heterogeneous vendorstorage arrays to be directly connected to its controllers. Theexternal storage assets presented to it are then discovered andmanaged in the same way as internal disks. This approach has a

number of benefits, including not requiring a remapping of LUNsor extents, meaning no additional layer of management and farless complexity in the network. Once virtualized in this manner,the sophisticated microcode software that resides on the storagecontroller presents the external storage assets as if they residedwithin the array, with hosts none the wiser of where they arephysically connected.



37/74


Figure 2-6: Controller-based storage virtualization allows external storage to

appear as if its internal.

This block-based virtualization effectively enables all

capabilities of the virtualized storage controller to be extendedto the external storage as well. Leveraging mature enterprise-

class features, data can be migrated non-disruptively from onepool to another, and replication can take place between non-like

and like storage. In addition, partitioning can be implementedto allocate resources such as ports, cache, and disk pools toparticular workloads to maintain quality of service and security,and thin provisioning can reduce storage allocation to other

assets. More recently, vendors have introduced clusteringcapabilities between controllers to provide the ultimate highavailability in the unlikely event of a complete storage array

failure.

Although a few downsides to controller-based virtualizationexist, the advantages not only far outweigh them but they also

address most of the deficiencies found in server- and network-based approaches.



38/74


Here are some of those advantages:

Connectivity to external storage assets is done via industrystandard protocols, with no proprietary lock-in.

Complexity is reduced as it needs no additional hardwareto extend the benefits of virtualization. In many cases therequirement for SAN hardware is greatly reduced.

Controller-based virtualization is typically cheaper thanother approaches due to the ability to leverage existingSAN infrastructure, and the opportunity to consolidatemanagement, replication, and availability tools.

Capabilities such as replication, partitioning, migration, andthin provisioning are extended to legacy storage arrays.

Heterogeneous data replication between non-like vendorsor different storage classes reduces data protection costs.

Interoperability issues are reduced as the virtualizedcontroller mimics a server connection to external storage.

All virtualization capabilities can be extended to other

forms of storage and platforms such as NAS, iSCSI, contentarchive, and virtual tape library through the connectionof appliance heads. Mainframes can also be connected,reducing islands of stranded storage. This enables onecommon management, data protection, and searchmethodology across the entire environment.

Interconnecting legacy assets and non-disruptive datamigration from one platform to another without anyapplication downtime assists in technology refresh cycles.

Infrastructure can be de-virtualized with full data integritysince no meta-data is maintained in appliances.

With virtualization embedded in the controller, anenterprise storage array has ten times the cache size ofa SAN controller and more port connectivity than a SAN-based virtualization appliance. This enables it to scale wellbeyond SAN-based appliances.

A storage controller can enhance the performance ofattached modular storage by as much as 30 percent. SAN-based virtualization cant enhance the performance ofexternal storage and in most cases will degrade it.

Vendors such as Hitachi Data Systems, Sun Microsystems, andHewlett Packard employ this form of storage virtualization.



39/74


Choosing the Right SolutionStorage virtualization is an enabler in your IT infrastructure;it helps you maximize the potential of storage assets, creatingefficiencies and automation, and helps avert risky situations. Buthow do you figure out which approach is the right solution foryou?

Consider the following issues and how they relate to yourenvironment to try to identify an appropriate storagevirtualization solution.

What are your objectives?Understanding the priorities of your business and how theIT environment can serve to deliver them is critical. If youvesuffered from application outage or data loss that had asignificant effect on your organization, exploring how storagetechnology can help avoid it happening again is the first place

to start. Or if you find the IT budget allocated by the business issimply insufficient to meet the growing data requirements, lookfor smarter ways of managing your storage resources. Take thetime to match these problems to specific capabilities offeredby storage virtualization. Ask a number of storage vendors toexplain how their solutions can meet your requirements.

Where are you today?Although the benefits of storage virtualization are enticing andcan make a tangible difference to your business, taking stockof where your organization is today in its level of maturity andcapability is important. After deciding storage virtualization canhelp you in your environment, consider your current capacityutilization and how much you can save by reallocating storageresources. Do the vendors you are talking to support the storagecomponents you already have in your environment? How matureare your storage management processes and do they need to be

addressed first? Do you have the in-house resources and skillsto undertake such an implementation, or will you need to hirecontractors or rely on your storage vendor? Do you have theright people to manage the environment after its all in place?



40/74


Price versus costImplementing storage virtualization needs to provide somemeasurable benefits to the organization; otherwise, lets face it,why would you do it in the first place? Ultimately, stepping intoany storage virtualization solution will require an investment oftime, people, and money. You will need to go through a discoveryphase, assign people to the transition, and possibly buy someequipment and/or software. However, remember that the price ofall this does not equal the cost. Implemented correctly, the rightstorage virtualization system, versus doing nothing, will realize

a return from the investment in efficiency savings. It is thereforeimportant to project what these savings will be so that you canevaluate which solution offers the best value. Not all vendorssolutions will be priced the same and some will offer greatercapabilities that lead to greater savings, so use your organizationsobjectives to prioritize the capabilities that are important. InChapter 3, we offer an approach for doing this.

Matching growth to scalabilityNot all organizations grow their storage at the same rate, so youneed to evaluate solutions that fit your growth profile. Althoughall storage vendors like to talk about the massive scalability andperformance their solutions provide, you need to cut throughthe marchitecture and only evaluate solutions that best suityour environment. Understand your real capacity utilization,your actual application growth (not storage allocation growth),and shortlist the capabilities you believe will make a difference,

perhaps thin provisioning and non-disruptive data migration.Insist on keeping to this scope.

Can you afford an outage?Every storage solution requires some disruption to yourorganization while it is implemented. Some solutions require agreater outage to your environment than others, depending on theamount of storage you have and the number and quality of skills

you have to carry it out.

As you evaluate solutions, understanding how much disruptionimplementing the solution will cause is important. You need tofeel comfortable that the storage or service vendor can minimizethis, not only through the technology itself, but through previousexperience.



41/74

Chapter 3

The Economics of StorageVirtualization

In This ChapterMeasuring the economic benefits of storage virtualization

Finding the savings in storage

Knowing where to start

Using Hitachis Storage Economics model

Looking at storage virtualization in action, a case study

When buying a new car, you dont just consider oneaspect, like the upfront cost of purchasing the car itself;you take into consideration multiple factors that determine theoverall value for money you are likely to get. You evaluate suchthings as how much mileage youre likely to get from a tankof gas, what the servicing and insurance costs are, and what

safety features are built in to protect you and your preciouspassengers.

The same is true for storage. If the cost of the capacity werethe only factor to consider, then the answer would be simple:Buy the disk from the guy who gave you the cheapest price perterabyte!

However, the impact of continuous growth in storage capacityon the IT budget can be complex, involving the interplay of adiverse set of costs. If all elements in the storage economicsequation arent taken into account, the real cost of a purchasingdecision can be decidedly different than intended.

In this chapter, we first look into these elements and just howyou can measure and evaluate them. Later in the chapter, weexplain Hitachis unique Storage Economics model, which



42/74


provides a methodology for measuring the benefits, andintroduce a case study to see how one organization realized

significant savings for its business.

How Do You Measure the EconomicBenefits?

Identifying cost categories that are important to the organizationis the first step in assessing the impact of a proposed storage

purchase. We detail these cost categories in the section WhereDo You Start? later in this chapter. Once identified, costs mustbe quantified using concise, deterministic methods such asreturn on investment (ROI), total cost of ownership (TCO), andreturn on assets (ROA), which we examine in this section. Othermethods of evaluation include time values of money saved,internal rates of return (IRR), and net present value (NPV) offuture savings. These metrics allow differences between new andold storage infrastructures to be impartially assessed.

Storage virtualization plays an important part in the storageeconomics equation as it is an enabling technology to accelerateefficiency and productivity improvements that justify a businesscase for change.

ROI and TCO are two important mechanisms used to determinethe economic value of a proposed purchasing decision. Althoughthe two are frequently treated as interchangeable in vendormarketing literature, ROI and TCO have very different purposes.

ROI is effective when challenging the status quo with aproposition to replace an existing solution. Frequently, ITplanners will have an idea that long-term operating expensesavings can be realized by making an investment, and the ROIanalysis provides the mechanism for fleshing out the financialpros and cons of the decision. The analysis illustrates how muchis to be invested, how quickly the investment is to be recouped,and what net savings are to be expected. Figure 3-1 shows a ROI

analysis over three years.



43/74

Chapter 3: The Economics of Storage Virtualization 33

Figure 3-1: This return on investment (ROI) over three years shows a break-

even point a little into the second year.

TCO analysis is used when the financial benefits of two or moreproposed solutions, or a solution against the status quo, mustbe assessed. The analysis places the total lifetime operatingand purchasing costs of the assets side by side for comparison(see Figure 3-2). IT purchasing best practices, documented

by independent analysts, frequently cite the importance ofincluding a TCO analysis in any competitive bid situation. Thecost categories chosen for comparison are those deemed mostsignificant, and these frequently vary from one organization tothe next.

Figure 3-2: Total cost of ownership (TCO), comparing the status quo with analternative.



44/74


However, ROA is a broader and more accurate measure of acompanys ability to make profit from its invested capital. ROA is

the ratio of net income to total assets. Many CxOs often regardIT storage as having a relatively poor ROA due to the historicallylow levels of capacity utilization caused by over-provisioning,data duplication, RAID (redundant array of independent disks)overhead, and stranded capacity. With the advent of storagevirtualization, and the ability to repurpose and reclaim legacyassets, thereby increasing their value, organizations can helprealize a faster return to the business, accelerate the savings,and reduce the TCO.

So Where Are the Savings inStorage?

For several years, Hitachi Data Systems has conducted exercisesfor hundreds of organizations around the world, assessing theeconomic impact of storage on their business. This work has led

to an astounding data point. On average, for every 12 terabytes(TB) of usable disk capacity within the storage infrastructure,there is potential for $1-million savings in net operating expense.

This single data point, which is often sufficient to justify furtherinvestigation and economic analysis, is reached through savingsacross the business environment. Here are the average savingsin each category:

Reduced waste, 25 percent

Reduced outages, 20 percent

Reduction in labor and management, 15 percent

Reduced maintenance fees, 15 percent

Environmental savings, 10 percent

Miscellaneous operating efficiency improvements, 5 percent

Other savings, 10 percent



45/74


Where Do You Start?The first step in assessing the impact of a storage system onthe IT budget in your business is to determine what expensecategories are real and defendable to the organization. Operatingexpenses (OPEX) fall into two primary categories: hard costs,the tangible and visible expenses that find their way intobudgets and financial reports; and soft costs, those less tangiblebut equally important measures, such as performance andavailability. Each organization will have its own idea of whichcosts are hard and which are soft.

Primary areas of investigation should include:

Data protection. Includes the cost of the organizationsbackup and business continuity infrastructure, whichmay include tape, replication, mirror sites, and all relatedplanning and operational activity.

Maintenance. The cost of hardware and softwaremaintenance is linked to the capacity deployed, whetherin use or not. It may increase as hardware warranties andbundled service deals expire.

Staff costs. The labor cost of storage administration andprovisioning can be significant, particularly in complexheterogeneous environments that may have multiplemanagement interfaces and distributed architectures.

Environment costs. Power, cooling, and floorspace costsare increasingly important as companies strive to meet

efficiency targets that are often driven as much by acorporate and social responsibility agenda as the needto save money. A significant number of companies arereaching or have reached absolute limits on their datacenter growth and power consumption.

Outage costs. These costs will vary widely depending onthe organization type, industry sector, and applicationconcerned. A realistic estimate of outage costs is a keyelement in planning the companys business continuity

arrangements.

Cost of growth. This will depend on the storage tieringstructure, since each tier is typically subject to differentgrowth rates, capital expenditure (CAPEX), and hardwareprice erosion. For example, lower tiers tend to grow morequickly than higher tiers and the price of low-cost SATA(serial advanced technology attachment) hard drives



46/74


47/74


1. Establish a TCO baseline. What does it cost theorganization to own and operate 1TB of storage per

year? The answer is derived from an understandingof the unique set of costs that apply to that storagecustomer, identified in up to 33 different cost areasranging from power and cooling to labor charges andoutage costs.

2. Map the costs to a set of strategic and tactical cost-saving initiatives. What cost-saving initiatives areneeded? These may be strategic or tactical initiatives,such as consolidation, tiered storage, virtualization,

thin provisioning, chargeback, or implementation ofmanagement tools, and will vary depending on thecustomers environment. The user maps the costsidentified in the first step to the cost-saving initiatives,so that the multiple relationships between costs andinvestment can be established. This process also helpsto identify the areas of technology investment that arethe highest priorities.

3. Compile a cost reduction roadmap. What will be the

impact on the TCO baseline over the short and mediumterms? With a rollout plan for each of the cost-savinginitiatives, a roadmap can be built that shows the impacton each of the key cost areas year by year.

By following this three-stage approach, a tactical plan foraddressing the key cost areas and driving long-term TCOreduction can be scheduled and executed.

Four principles of storage economicsHitachi has developed an online

tutorial outlining four principles toreduce the total cost of ownership(TCO) with storage virtualization.

Go to www.hds.com/economize.



48/74


Case study: Indosats storage solutionIndonesian Satellite Corporation Tbk(Indosat), one of Indonesias largest

telecommunications providers, wasformed with the merger of three separ-ate entities. The merger transformed

the company into a cellular-focused,full-network service provider ser-vicing over 12 million customers, and

presented a major challenge.

Working with three different sets ofstorage systems threatened to takea toll on Indosats IT infrastructureoperations and maintenance. It wasclear that direct attached storage foreach server was no longer an optionas it would not meet Indosats require-ments for scalability and flexibility.

In addition to this challenge, a rapidlygrowing customer base was straining

the capacity of the existing systems.A disaster recovery facility was alsoneeded to protect critical applicationdata in the event of systems failures.Consolidation of the IT infrastructures,as well as the applications, becameessential.

Indosat identified Hitachi DataSystems as the solutions provider ofchoice to quickly take control of thesituation. By drawing on a combi-nation of skills in consolidatingstorage resources and leveraging

the unique virtualization capabilitiesof the Universal Storage Platform,Hitachi Data Systems designed a solu-

tion that integrated the three storageinfrastructures seamlessly into oneenvironment while leveraging existinginvestments. Indosats applicationswere classified into relevant tiers

that matched the value of the data to the cost of the storage, ensuring theorganization was confident they had asustainable way of optimizing storageresources that raised efficiency andaccountability.

Virtualization paved the way for con-solidation and allowed Indosat tomake the best use of their existingstorage, effectively protecting thebot

Hitachi Storage Virtualization

Documents

Transcript of Hitachi Storage Virtualization